Skip to content

CHANGELOG.md

Latest commit

 

History

History
161 lines (119 loc) · 8.03 KB

CHANGELOG.md

File metadata and controls

161 lines (119 loc) · 8.03 KB
Raw

Changelog

[0.1.7] - 2026-08-04

Features

  • QA Doris read query: POST /webapi/api/v1/ai/doris/query and built-in tool platform.queryDoris so the product QA expert can verify platform config and telemetry via the web JDBC pool (read-only, row-capped)
  • QA skill / management APIs: product QA skill documents queryDoris plus frontend management API login/credential discovery for config diagnosis

Bug Fixes

  • Fix portal resourceInfo / resourceRelation for non-HTTP componentTypes
  • Honor lt / lte / gte comparators and warning thresholds in monitor rule evaluation
  • Apply remote MCP headers after transport; add Streamable HTTP integration coverage
  • Fix AI chat Mermaid rendering (per-node with syntax-repair fallback) and tighten expert Mermaid prompt hints
  • Fix getCurrentTimeRange end-time handling; remove unused LLM max_token default

Documentation

  • Update download URL and related install docs

Deploy & Build

  • Docker / K8s / offline install scripts versioned to 0.1.7

Full changelog

Commits since 0.1.6:

  • 2890b41 Add read-only Doris query API and queryDoris tool so the QA expert can verify platform config and data without ad-hoc DB connections
  • 08fc79e fix(portal): support non-HTTP componentTypes in resourceInfo and resourceRelation
  • eeb246e update docs
  • 8cde082 fix(monitor): honor lt/lte/gte comparators and warning thresholds in rule evaluation
  • 0ff8ae2 fix(mcp): apply remote MCP headers after transport and add Streamable HTTP IT
  • 1d5f0a3 update getCurrentTimeRange for end
  • 853552b update download url
  • f67a2b0 fix(ai-chat): render mermaid per-node with syntax-repair fallback and tighten expert mermaid prompt hints
  • 7b347bc remove llm max_token default value

[0.1.6] - 2026-07-30

[0.1.5] - 2026-07-27

Features

  • OTLP compression: ingest supports gzip / snappy / zstd (and HTTP zlib / deflate / lz4) on OTLP trace and metric payloads
  • Span ignore lists: drop spans by resource exact or regex ignore rules before fill and metrics
  • Mermaid in AI chat: experts can render topology and flow diagrams inline; prompts steer experts toward Mermaid for structure
  • Log detail drawer: right-side drawer loads full log fields on demand from the log list
  • Size-based Doris flush: thread-local buffers and size-triggered handoff reduce lock contention on high-throughput ingest

Improvements

  • Doris write tuning: default flush batch raised to 50 MiB; telemetry dynamic-partition buckets reduced from 16 to 3 (V005)
  • HTTP span URL normalization: unify resource-detail HTTP span queries on path-only url; normalize meta.http.url at ingest
  • LLM config resilience: reload LLM settings from Doris on every probe so keys survive full-server restarts
  • JVM safety: add -XX:+ExitOnOutOfMemoryError to ingest / web runtime images
  • AI session stability: pending tool recovery for all session-scoped experts; Doris hydrate no longer tears down live expert runtimes; Mermaid rendering stabilized
  • AI test parallelism: AI suites and cases run in parallel with same-session parallel fan-in expectations

Bug Fixes

  • Fix service instance metadata mapping edge cases
  • Fix Anthropic tool-call formatter and align tool-call content with input via middleware
  • Fix trace query error messaging when Doris tablet version graph is broken

Documentation

  • Migration guides and competitor comparison docs (Jaeger, SigNoz, SkyWalking, Pinpoint, OpenObserve)
  • Product intro and README updates; blog content

Deploy & Build

  • Docker / K8s / offline install scripts versioned to 0.1.5
  • Schema migration V005: reduce dynamic-partition telemetry table buckets to 3

Full changelog

Commits since 0.1.4:

  • 35d3106 update docs
  • 3904f63 fix(ai): stabilize Mermaid chat rendering and prompt experts to use it for topology
  • 1429037 fix(ai): stop Doris hydrate from tearing down live expert session runtimes
  • d29dc35 update doris flush max size to 50M
  • 6cabe39 update table buckets to 3
  • edcd4b6 test(ai): run AI suites and cases in parallel
  • ad5b2da fix service instance meta bug
  • 46d7ec0 feat(ingest): size-based Doris flush with thread-local buffers
  • 54fe499 add -XX:+ExitOnOutOfMemoryError
  • 90c33a2 feat(ingest): drop spans by resource ignore lists
  • ca7f53a feat(ingest): support OTLP gzip/snappy/zstd compression
  • aa7045f fix: reload LLM config from Doris on every probe
  • b907835 Add log detail drawer with on-demand full fields
  • 14c8d3c fix(ai): align tool-call content with input via middleware
  • f0f768b support Mermaid
  • 53063b2 fix(ai): enable pending tool recovery for session-scoped experts
  • da56fa6 fix(resource-detail): unify HTTP span queries on path-only url
  • f114392 update to v0.1.5

[0.1.4] - 2026-07-15

Features

  • SkyWalking native ingest: normalize SkyWalking/OTel MQ producers into virtual services with messaging tags and producer outbound routing (#34)
  • OTel span/resource attributes merge: merge span and resource attributes into a single map in buildDcSpan for unified attribute access
  • Doris runtime failover E2E: new comprehensive test suite (deploy/test/doris-runtime-failover-e2e.sh + Python ops chat) — release gate B
  • Persistence startup hydrator: re-hydrate persistence after Doris recovery from troubleshooting mode

Improvements

  • Doris VARCHAR limits: expand long-text VARCHAR column limits and add ingest-side truncation to prevent schema violations (#38)
  • Doris history partitions: create historical partitions in V003 migration so metric backfill no longer fails on missing partitions
  • Doris availability monitoring: refactor DorisAvailabilityMonitor for faster fail-detection and recovery cycle
  • Docker compose: drop mem_limit for docker-compose 1.22 / Compose file v3 compatibility; remove CPU limits from compose files
  • Legacy compose: add docker-compose.legacy.yml for older Docker Compose versions
  • Runtime script: add runtime.sh with ensure_vm_max_map_count helper for Docker deployments
  • RPC call detail: hide always-empty Thread Name columns in RPC call detail table
  • Doris failover E2E: enhance with configurable STOP_AFTER and improved assertions

Bug Fixes

  • Fix Doris JDBC gate to fail-close before web port opens (prevents 5s API hang when Doris is unreachable)
  • Fix OTel metric row mapping edge cases in OtlpMetricRowMapper
  • Fix OptimizedMetricAccumulator accumulation logic
  • Fix demo compose file compatibility

Documentation

  • Python OTLP integration quick-start guide (CN + EN)
  • Webhook notification configuration guide (CN + EN)
  • Docker ops guide updates for release gates (CN + EN)
  • Updated README_en.md

Deploy & Build

  • Docker images versioned to 0.1.4
  • K8s install/download scripts updated for 0.1.4
  • Offline install/update/demo scripts updated
  • build-docker.sh packages SQL migrations and upgrade-manifest.json

Full changelog

Commits since 0.1.3:

  • 2926894 update version to 0.1.4
  • 41f1bd6 fix: drop compose mem_limit for docker-compose 1.22 / Compose file v3 compatibility
  • dbde89f fix: expand Doris long-text VARCHAR limits and truncate on ingest (#38)
  • 5b75ab8 Merge otel span/resource attributes into a single map in buildDcSpan
  • cfb0590 去掉docker cpu limit
  • 76ac2ca normalize skywalking/otel mq producers into virtual services like redis with messaging tags and producer outbound
  • 2529aee Merge pull request #34 from kiddo90-N/master
  • e4d4f9c docs: add python otlp integration quick-start guides
  • ee4c128 docs: integrate webhook configuration guide and update notification status
  • 773574c Update README_en.md
  • 6046bee docs: add English configuration guide for webhook notification channel
  • c3ad7c9 fix: create Doris history partitions in V003 so metric backfill no longer fails
  • ad3a73b Merge pull request #25 from mvanhorn/fix/rpc-hide-empty-threadname-columns
  • 7aa39e1 fix: hide always-empty Thread Name columns in RPC call detail table
  • 1ec7794 update version to v0.1.4

Contributors: Brio Griondy Dahlinar, Matt Van Horn, databufflabs, ligang